Scholarlink

<p>Powder-based additive manufacturing processes, particularly binder jetting followed by sintering, offer significant advantages for producing metal components with complex geometries and high material efficiency. However, these techniques inherently introduce microstructural anisotropy due to directional effects from layer-by-layer deposition and thermal gradients during sintering. Accurate quantification of key microstructural features—specifically porosity, particle contact area, and particle contact orientation—is essential for understanding and mitigating this anisotropy to enhance mechanical performance. This study presents a machine learning (ML) framework for the automated characterization of pores and inter-particle contact features in binder-jetted and sintered stainless-steel components using three-dimensional X-ray computed tomography data. The framework integrates advanced image processing with deep learning models to extract quantitative descriptors of porosity, contact area, and contact orientation. It is applied to analyze anisotropic microstructures in samples produced under three different sintering protocols. The results demonstrate the ML framework’s effectiveness in quantifying complex, direction-dependent microstructural features in additively manufactured metal parts. By enabling accurate, scalable, and non-destructive analysis, the proposed method offers valuable insights for optimizing process parameters and enhancing the structural reliability of metal components produced by powder-based additive manufacturing.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-driven characterization of anisotropic microstructures in powder-based additive manufactured metal parts

  • 期刊论文

  • 发表日期：

    2026年7月20日

查看全文
  • Nicholas Satterlee,
  • Runjian Jiang,
  • Elisa Torresani,
  • Thomas Grippi,
  • Andril Maximenko,
  • Marc P. F. H. L. van Maris,
  • Diletta Giuntini,
  • Eugene Olevsky,
  • John S. Kang

摘要

Powder-based additive manufacturing processes, particularly binder jetting followed by sintering, offer significant advantages for producing metal components with complex geometries and high material efficiency. However, these techniques inherently introduce microstructural anisotropy due to directional effects from layer-by-layer deposition and thermal gradients during sintering. Accurate quantification of key microstructural features—specifically porosity, particle contact area, and particle contact orientation—is essential for understanding and mitigating this anisotropy to enhance mechanical performance. This study presents a machine learning (ML) framework for the automated characterization of pores and inter-particle contact features in binder-jetted and sintered stainless-steel components using three-dimensional X-ray computed tomography data. The framework integrates advanced image processing with deep learning models to extract quantitative descriptors of porosity, contact area, and contact orientation. It is applied to analyze anisotropic microstructures in samples produced under three different sintering protocols. The results demonstrate the ML framework’s effectiveness in quantifying complex, direction-dependent microstructural features in additively manufactured metal parts. By enabling accurate, scalable, and non-destructive analysis, the proposed method offers valuable insights for optimizing process parameters and enhancing the structural reliability of metal components produced by powder-based additive manufacturing.